Mania Africa 💎🌍

Geekline Feed Archive

113 posts
New from 404 Media: companies are deliberately making Claude and Codex talk like cavemen because it can curb the skyrocketing cost of AI. The 'caveman' plugin keeps LLM answers very, very brief. Got leaked memo showing it's being used; OpenAI dev contributed code too www.404media.co/companies-ar...
Media from Joseph Cox
View Original BlueSky
[Image]
Media from Tom's Hardware: For The Hardcore PC Enthusiast [Unofficial]
View Original BlueSky
Google DeepMind 𝕏🔁 @IasonGabriel@twitter.com:

What does it mean to be an ethicist and a philosopher working in an AI company?

What questions do we explore – and can we affect positive change?

For a deep exploration of these themes, check out a new long-read […]

[Original post on zpravobot.news]
Media from Google DeepMind :bot:
View Original BlueSky
💜🤝🌱 Aikido acquires Root.

Uniting behind one mission: make it easy to build with secure open source and take on supply chain attacks.

Read more: www.aikido.dev/blog/aikido-...
View Original BlueSky
Linux 7.2 Features: Cache Aware Scheduling, USB4STREAM, AMD ISP4, AMDGPU HDMI 2.1 FRL -
View Original BlueSky
Official support for Windows 10 has ended, and its security certificates are expiring. Your system is at risk, and Microsoft wants you to upgrade. But an extra year of extended support can grant you some extra time.
Media from
View Original BlueSky
Schneider Electric agrees to buy industrial AI company Cognite in a $3.1B all-cash deal, and plans to combine Cognite with its industrial software company Aveva (Frank Connelly/Bloomberg)

Main Link | Techmeme Permalink
View Original BlueSky
OT/ICS Cybersecurity Career Roadmap: Breaking Down the 7 Specializations That Will Define Critical Infrastructure Defense + Video

Introduction The convergence of operational technology (OT) and information technology (IT) has created a critical skills gap in industrial cybersecurity, with…
Media from Undercode Testing
View Original BlueSky
While Nero plays the fiddle-
A key Department of Homeland Security information-sharing database was accessed by an unknown threat actor in recent weeks, potentially exposing sensitive data exchanged between federal, state, local and industry partners, according to two people familiar with the matter
Media from David blr
View Original BlueSky
According to Ransomware.live, chaos ransomware group has added universalplant.com (🇺🇸) to its victims.
View Original BlueSky
New cybersecurity guidelines have been issued for the safe use of artificial intelligence by the public service
Media from RTÉ News
View Original BlueSky
A bot's code got rejected from matplotlib, so it wrote a hit piece on the human who rejected it — and no one could find the person behind it. 'Proof of personhood' is one proposed fix -- but can it be done without a checkpointed internet?

My new essay up at Noema: www.noemamag.com/we-need-a-wa...
Media from Renee DiResta
View Original BlueSky
OpenAI unveils Genebench-Pro, a new benchmark with 10 real-world genomics case studies. It pushes AI models to solve complex genetic and clinical problems, demanding precise analysis.
Media from StartupHub AI
View Original BlueSky
Hundreds of contractors used dummy under-18 accounts to probe ChatGPT, Gemini, and Character.AI with safety-evasion prompts and logged replies without the targets’ knowledge.
Media from Privacy Professionals Briefly
View Original BlueSky
Media from Media & Communications Law
View Original BlueSky
OpenAI's bug hunt: Engineers used 'epidemiology' to find an 18-year-old GNU libunwind flaw and a hardware issue causing crashes in their data infrastructure.
Media from StartupHub AI
View Original BlueSky
AWS is committing $1 billion to embed AI engineers inside customer businesses:
View Original BlueSky
Google rolls out Nano Banana 2 Lite for rapid image generation and Gemini Omni Flash for high-quality video editing, enhancing developer capabilities in generative AI.
Media from StartupHub AI
View Original BlueSky
Yesterday, a message appeared on DCHA computers claiming attackers accessed sensitive information and threatened to release it within 72 hours, according to a screenshot shared with us.

The allegations have not been independently verified, but DCHA is working to restore operations. Full story:
Media from Washington City Paper
View Original BlueSky
Anthropic unveils 'Claude Science' AI platform for scientific research -
Media from CNA - Channel News Asia (Unofficial)
View Original BlueSky
Google will shut down the Tenor API, forcing apps that used it to migrate to other GIF picker services.
Media from Digital Life Briefly
View Original BlueSky
Media from How-To Geek
View Original BlueSky
📰 #Finance

"More than 200 insurance leaders gathered at Earnix’s North American headquarters for the introduction of Earnix AIOS, the launch of its first Customer Advisory Board, and a keynote from Professor Kevin Fong on high-stakes decision-making BOSTON — Earnix, the first AI…"
View Original BlueSky
9to5Mac より

Cibby is a beautifully designed iOS app for physical video game collectors
Cibbyは物理的なビデオゲームコレクターのための美しいデザインのiOSアプリです
Media from gigowat Ver.3.0
View Original BlueSky
Meta releases version two of its brain-computer interface that can turn thoughts into keypresses — non-invasive magnetoencephalography scanner can measure changes in brain activity

www.tomshardware.com/peripherals/...
Media from TruthPhoenix
View Original BlueSky
Source: OpenAI engineers earlier this month told some colleagues they had figured out a way to more than halve the cost of inference (Stephanie Palazzolo/The Information)

Main Link | Techmeme Permalink
View Original BlueSky
🟢 Law enforcement disrupt operations of the Amadey and StealC malware infrastructures

🗨️ As part of the international operation Endgame, law enforcement agencies and cybersecurity specialists from various comp…

#news
Media from HackMag — Top-notch cybersecurity magazine
View Original BlueSky
Validin upgraded its data pipeline to catch short-lived infrastructure, moved key DNS scans to daily, and added Bulk Search to enrich up to 1,000 indicators. Threat Profiles were redesigned for faster analysis. #Validin #DNS #BulkSearch
Media from Cybersecurity News Everyday
View Original BlueSky
Framingham DPW (via Facebook): Hello, July! Summer is in full swing, and Framingham Parks and Recreation has a packed schedule of fun, fitness, and community events to keep you and your family busy all month long! From weekly sports clinics and Storytime in the Parks to live summ… [Link] #Framingham
Media from Framingham City Posts Bot - UNOFFICIAL!
View Original BlueSky
A study of 22K US companies shows those spending most heavily on AI are adding workers faster than peers, but most gains are among tech companies and startups (Financial Times)

Main Link | Techmeme Permalink
View Original BlueSky
Anthropic launches Claude Sonnet 5 as a cheaper way to run agents
techcrunch | Author: Rebecca Bellan
Media from Cartero
View Original BlueSky
ESMO GI 2026: Unlocking the Power of Oncology Data with AI and Cybersecurity + Video

Introduction The European Society for Medical Oncology (ESMO) Gastrointestinal Cancers 2026 conference (ESMOGI26) serves as a pivotal gathering for oncologists, researchers, and data scientists to present…
Media from Undercode Testing
View Original BlueSky
JSP webshells being dropped on unpatched PTC Windchill instances

The US Cybersecurity and Infrastructure Security Agency (CISA) added a vulnerability (CVE-2026-12569) in Windchill and FlexPLM, two product lifecycle management software platforms developed by PTC, to its Known Expl…
#hackernews #news
View Original BlueSky
The initiatives coincide with public debates about South Korean chipmakers’ huge profits during the AI boom, as well as proposals by policymakers to distribute that excess wealth.
Media from Ars Technica
View Original BlueSky
Halving inference costs is a rather big deal.

"OpenAI applied the new optimizations to ChatGPT, specifically for visitors who don't have an account. The number of Nvidia GPUs needed to serve those users dropped to just a few hundred"
Media from Claes Holtzmann
View Original BlueSky
Probe-based identification of metal-binding sites using deep learning representations @natcomms.nature.com
View Original BlueSky
EDR Black Box No More: How LLMs Are Democratizing Evasion and Reshaping Endpoint Security + Video

Introduction For years, Endpoint Detection and Response (EDR) solutions have been perceived as impenetrable black boxes—observing, correlating, and detecting while attackers operated in the dark,…
Media from Undercode Testing
View Original BlueSky
In the ongoing storyline over if Ai companies are going to pay for the content they use, Open AI has struck a deal with Getty images

qz.com/getty-images...
Media from Clare Spencer
View Original BlueSky
Anthropic PBC is releasing new software aimed at helping scientists automate research, reducing some tedious aspects of the work.Read here financialpost.com/technology/a...
Media from Financial Post
View Original BlueSky
I've had fascinating discussions over the past week with CISOs and security leaders about how they vary which AI model to use--and they use multiple models--depending on the complexity of the work to save on cost and time.

Anthropic debuts Sonnet 5 for everyday work
www.axios.com/2026/06/30/a...
Media from Cynthia Brumfield
View Original BlueSky
Claude Sonnet 5 costs $2 per 1M input tokens and $10 per 1M output tokens through August 31, after which prices rise to $3 and $15, respectively (Zac Hall/9to5Mac)

Main Link | Techmeme Permalink
View Original BlueSky
Anthropic has renegotiated part of its Amazon partnership, shifting pricing toward tokens instead of computing hours.

The change could raise Amazon’s costs as it uses Claude across shopping, coding and workplace AI products.

Read more:
Media from The Information
View Original BlueSky
Mastering Digital Forensics and Incident Response: The Ultimate Threat Hunting Training for 2026 + Video

Introduction: In an era where cyber adversaries are increasingly sophisticated, the ability to swiftly detect, analyze, and respond to security incidents has become paramount for organizations…
Media from Undercode Testing
View Original BlueSky
Meta hired hundreds of contractors who posed as teens to secretly test ChatGPT, Gemini, and Character.AI on topics like suicide, sex, and drugs. The project was active as of April 2026. Meta calls it safety testing. Others might call it something else entirely.

newslit.substack.com/p/cma-hits-a...
Media from Newslit News
View Original BlueSky
[Image]
Media from Ars Technica
View Original BlueSky
After a dramatic ban on Anthropic’s AI for foreign use, which sent shockwaves across the tech industry, the US government has partially eased restrictions on Anthropic's advanced Mythos 5 model following negotiations aimed at strengthening cybersecurity safeguards.
Media from Idenhaus
View Original BlueSky
If you've been wondering about the NIST's reduction in CVE enrichment and the effects the shift has had since it was implemented two months ago, the folks at Volerion analyzed more than 13,000 CVEs published to the NVD and found some concerning trends. www.darkreading.com/vulnerabilit...
Media from Rob Wright
View Original BlueSky
meta paid contractors to pose as kids, testing how far chatgpt and gemini would go on suicide and sex
then called it routine safety testing
Media from ai-nerd.bsky.social
View Original BlueSky
Event tomorrow!

Join CSIS Strategic Technologies Program's Dep. Director Lauryn Williams and Adjunct Fellow Carol Kuntz tomorrow, July 1, for a discussion on the growing role of artificial intelligence in cyber defense.

Register here:
Media from CSIS Economic Security and Technology Department
View Original BlueSky
I dunno who needs it but this was an amusing (in a dark way). nesbitt.io/2026/06/26/i...

I think my favorite line is from the customer impact section: “Some customers may have experienced unscheduled collaborative compute with external parties.” 💀

#aisecurity
#weredoomed
Media from Ed Weadon
View Original BlueSky
View Original BlueSky
A large number of GPT-5.6 hands-on tests have leaked: it may debut this Thursday

OpenAI's kindle-alpha and GPT-Bidi-1 are nearly here. Here's what the leaks, routing logs, and early demos actually tell us — and what's still speculation.
#gpt #hackernews #openai
View Original BlueSky
Your Space is My Zone: Demystifying the Security Risks of AI-Powered Applications on Pre-Trained Model Hubs

Yacong Gu et al.

#arXiv #cs.CR
View Original BlueSky
According to Ransomware.live, play ransomware group has added Western Construction (🇺🇸) to its victims.
View Original BlueSky
The White House mandated that federal agencies transition to quantum-safe cryptography by 2031 to counter adversary data harvesting. In ClearanceJobs, Doc McConnell explains why cryptographic transparency & supplier evidence are the core requirements for meeting the deadline. https://bit.ly/4oPqvIB
Media from Finite State
View Original BlueSky
Google launches Nano Banana 2 Lite and Gemini Omni Flash. How to try them now

Google has some new AI tools for users to mess around with. The company announced in a blog post on Tuesday that Nano Banana 2 Lite and Gemini Omni Flash are now available. These are a couple of new toys that live within…
Media from
View Original BlueSky
Anthropic just launched Claude Science, a dedicated science desktop app that helps researchers do their work. #Anthropic #ClaudeScience
Media from Neowin
View Original BlueSky
Google shuts down the Tenor API, affecting GIF pickers on platforms like Discord, WhatsApp, and Bluesky; Nikita Bier says X has migrated elsewhere (Ben Schoon/9to5Google)

Main Link | Techmeme Permalink
View Original BlueSky
I ask again, where are all the techno-optimists who breathlessly warned about bureaucrats impeding progress and equated AI deceleration with murder? Isn't the Anthropic ban *exactly* the kind of red line they swore to oppose? What happened?
View Original BlueSky
Also Anthropic horked their API for Sonnet 5, not bothering to update their documentation when doing so

I’m really thinking about just giving Anthropic users a refund of their full purchase price, no provider causes as much trouble as whatever PM!Claude does to itself
View Original BlueSky
Avoid Installing These Specific Android Apps Right Now – Uninstall If you have any of these Apps.

Google’s Android Feature Drop added better scam protection, but dangerous apps are still flooding the Play Store and third-party app stores sources. In this updated guide, I name specific apps and app…
Media from Paddedvibez
View Original BlueSky
Daily Tech Headlines for Tuesday June 30, 2026 with Robb Dunewood
Supreme Court Rules Search Warrant Required for Geofence Location Data
Listen here: shows.acast.com/dth/episodes...
Media from Daily Tech News Show
View Original BlueSky
yeah, i need to finish moving off anthropic for personal use this week.

(i am subject to my employer’s policies for work usage, which i will not discuss in detail due to reasons.)
View Original BlueSky